Re: [patch 2.6.15-current] i386: multi-column stack backtraces

[Dave Jones <davej@redhat.com>]

On Mon, Jan 16, 2006 at 10:42:34PM -0800, Andrew Morton wrote:

 > Presumably this is going to bust ksymoops.
 
Do people actually still use ksymoops for 2.6 kernels ?

I resorted to it about 6 months ago for the first time in the
better part of 3 years, and it didn't even compile.
(I only wanted to disassemble a Code: line, addr resolution
 works for me [and most other distros afaik] with kksymoops now)

 > Also the various other custom
 > oops-parsers which people have written themselves.

Given we've extended the oops output in several different
ways without thought in the past, it seems a bit late.
We added printing of module list in the middle of the output.
We added various tainting flags over time.

What other tools parse oopses ? ksymoops is the only one I recall.

 > The patch is a desirable change (I do get seasick reading x86_64 traces,
 > but I'll get over it), but it'll cause various bits of downstream grief.

I'd be surprised if anyone noticed.

*shrug*, in an ideal world, no-one would ever see an oops anyway :-P
 
		Dave